Predicting fault slip via transfer learning

Generative Artificial Intelligence GPT-4 Accelerates Knowledge Mining and Machine Learning for Synthetic Biology

Knowledge mining from synthetic biology journal articles for machine learning (ML) applications is a labor-intensive process. The development of natural language processing (NLP) tools, such as GPT-4, can accelerate the extraction of published information related to microbial performance under complex strain engineering and bioreactor conditions. As a proof of concept, we proposed prompt engineering for a GPT-4 workflow pipeline to extract knowledge from 176 publications on two oleaginous yeasts (Yarrowia lipolytica and Rhodosporidium toruloides). After human intervention, the pipeline obtained a total of 2037 data instances. The structured data sets and feature selections enabled ML approaches (e.g., a random forest model) to predict Yarrowia fermentation titers with decent accuracy (R2 of 0.86 for unseen test data). Via transfer learning, the trained model could assess the production potential of the engineered nonconventional yeast, R. toruloides, for which there are fewer published reports. This work demonstrated the potential of generative artificial intelligence to streamline information extraction from research articles, thereby facilitating fermentation predictions and biomanufacturing development.

Machine learning analysis and risk prediction of weather-sensitive mortality related to cardiovascular disease during summer in Tokyo, Japan

Climate-sensitive diseases developing from heat or cold stress threaten human health. Therefore, the future health risk induced by climate change and the aging of society need to be assessed. We developed a prediction model for mortality due to cardiovascular diseases such as myocardial infarction and cerebral infarction, which are weather or climate sensitive, using machine learning (ML) techniques. We evaluated the daily mortality of ischaemic heart disease (IHD) and cerebrovascular disease (CEV) in Tokyo and Osaka City, Japan, during summer. The significance of delayed effects of daily maximum temperature and other weather elements on mortality was previously demonstrated using a distributed lag nonlinear model. We conducted ML by a LightGBM algorithm that included specified lag days, with several temperature- and air pressure-related elements, to assess the respective mortality risks for IHD and CEV, based on training and test data for summer 2010-2019. These models were used to evaluate the effect of climate change on the risk for IHD mortality in Tokyo by applying transfer learning (TL). ML with TL predicted that the daily IHD mortality risk in Tokyo would averagely increase by 29% and 35% at the 95th and 99th percentiles, respectively, using a high-level warming-climate scenario in 2045-2055, compared to the risk simulated using ML in 2009-2019.

On catching the preparatory phase of damaging earthquakes: an example from central Italy

How, when and where large earthquakes are generated remain fundamental unsolved scientific questions. Intercepting when a fault system starts deviating from its steady behavior by monitoring the spatio-temporal evolution and dynamic source properties of micro-to-small earthquakes can have high potential as tool for identifying the preparatory phase of large earthquakes. We analyze the seismic activity that preceded the Mw 6.3 earthquake that hit L'Aquila on 6 April 2009 in central Italy, and we show that the seismic catalog information can be transformed into features allowing us to track in a statistical framework the spatio-temporal evolution of seismicity. Features associated to foreshocks show different patterns from the background seismicity that occurred in the previous years. We show that features ensemble allows to clearly capture the activation phase of the main event. Nonetheless, foreshocks share similar clustering properties of previous seismic sequences not culminating in large earthquakes, and thus generating questions on their use as potential precursor for earthquake sequences prone to evolve into catastrophic sequences.

Using a physics-informed neural network and fault zone acoustic monitoring to predict lab earthquakes

Predicting failure in solids has broad applications including earthquake prediction which remains an unattainable goal. However, recent machine learning work shows that laboratory earthquakes can be predicted using micro-failure events and temporal evolution of fault zone elastic properties. Remarkably, these results come from purely data-driven models trained with large datasets. Such data are equivalent to centuries of fault motion rendering application to tectonic faulting unclear. In addition, the underlying physics of such predictions is poorly understood. Here, we address scalability using a novel Physics-Informed Neural Network (PINN). Our model encodes fault physics in the deep learning loss function using time-lapse ultrasonic data. PINN models outperform data-driven models and significantly improve transfer learning for small training datasets and conditions outside those used in training. Our work suggests that PINN offers a promising path for machine learning-based failure prediction and, ultimately for improving our understanding of earthquake physics and prediction.